Grasp Rehabilitation of Stroke Patients Through Object Manipulation with an Intelligent Power Assist Robotic System

Abstract

A 1-DOF experimental power assist robotic system was developed for grasping and lifting objects tied with it by stroke patients as a part of their upper arm grasp rehabilitation practices. A human-in-the-loop model of the targeted dynamics for lifting objects with the system was developed that considered prospective patient’s weight perception. A position control scheme based on the weight-perception-based dynamics model was developed and implemented. The control parameters were the virtual mass values used for the grasped and manipulated object. The virtual mass values that produced satisfactory human-robot interactions (HRI) in terms of maneuverability, motion, stability, health and safety, etc. were determined. The results showed that a few sets of mass values could produce satisfactory HRI. Then, stroke patients grasped and manipulated objects with the system separately for the most satisfactory virtual mass values that provided optimum kinesthetic perception. In separate experiments, the virtual mass value was exponentially reduced when the patients manipulated the objects with the system, which provided variation in kinesthetic perception. The results showed that the power-assisted manipulation with optimum kinesthetic perception significantly contributed to stroke rehabilitation. The results also showed that the rehabilitation performance varied with the variation in the kinesthetic perception. In addition, effectiveness of a proposed game-like visual interface and smart tele-rehabilitation for connected health was investigated.